Chaos characteristics and its application to prediction of runoff time series

Abstract

In the recent years, many researchers in different fields often apply the stochastic and nonlinear theories to study data time series. The stochastic and nonlinear methods of analyzing hydrological data such as discharge, water stage, sediment yield are different from the traditional deterministic methods. In this study, the Fractal Dimension, Lyapunov and Kolmogorov index etc. For the daily runoff time series have been analyzed, the preliminary results appear that the daily runoff time series has chaotic characteristics. Further, predictions are made by applying the local approximation method to daily runoff time series, from the comparison between the predictions and observations of daily runoff data, the daily runoff time series can be convincingly modeled by the time delay embedding approach of the chaotic theory.